Nebius delivers Europe’s first live NVIDIA GB300 NVL72 deployment
Nebius delivers Europe’s first live NVIDIA GB300 NVL72 deployment
In November, we announced the arrival of NVIDIA Blackwell Ultra AI infrastructure to the UK. Building on that announcement, Nebius’ expanded Finland data center now hosts Europe’s first operational production deployment of the NVIDIA GB300 NVL72 platform, expanding the capabilities of Nebius AI Cloud.
Powered by NVIDIA Blackwell Ultra GPUs and NVIDIA Quantum-X800 InfiniBand networking, the system brings unprecedented performance and system-level efficiency for generative AI, large-scale model training, and advanced inference workloads — and represents Europe’s first GB300 NVL72 system running on NVIDIA Quantum-X800 InfiniBand.
With 72 Blackwell Ultra GPUs operating as a single, tightly integrated supercomputer node, NVIDIA GB300 NVL72 integrates high-bandwidth, low-latency NVIDIA NVLink 5 interconnects to enable unified GPU memory access and deliver rack-scale performance that behaves like a single, giant GPU for large-scale AI and HPC workloads. Compared to previous NVIDIA systems, GB300 NVL72 delivers an impressive 1,000 tokens per second, 50x boost in AI Factory output performance and provides a 30x performance improvement for real-time video generation.
By pairing the supercomputer-class architecture of NVIDIA GB300 NVL72 with the scalability, elasticity, and the native and ecosystem AI/ML tooling that Nebius AI Cloud offers, customers can train, fine-tune and serve even the most demanding AI models without bottlenecks.
The new racks, featured below, were showcased during a recent tour of the data center led by CEO Arkady Volozh, where guests got a sneak preview of the now-operational NVIDIA GB300 systems.
Photo 1. Nebius’ Co-Founder and CEO Arkady Volozh at our own data center in Finland
Photo 2. Kelly Kung, Investor at Accel, and Nebius’ Co-Founder and CEO Arkady Volozh
Photo 3. Server racks in Nebius’ data center in Finland
Photo 4. NVIDIA Blackwell systems deployment in Nebius’ data center in Finland
With this rollout, Nebius expands the compute foundation available to Europe’s AI builders — enabling the next generation of foundation models, agentic systems and sovereign AI initiatives.
Learn more about our latest release here.
Explore Nebius Token Factory
See also
The Nebius AI Cloud 3.1 release continues the evolution of our platform into a transparent, orchestrated, developer-first environment — where performance is accessible, capacity is visible and control feels effortless. Across new platforms based on NVIDIA Blackwell Ultra, smarter capacity management and an improved experience for both developers and platform operators, every enhancement moves us closer to that goal.
In this blog, we will take a look at how teams can deploy the four NVIDIA NIMs available now in our platform, how they can be consumed via their API and how they can fit into larger scientific workflows. It also covers the Virtual Screening Blueprint, which turns folding, molecule generation and docking into a ready workflow that becomes available in JupyterHub shortly after deployment.
Today, we’re expanding Nebius AI Cloud’s elasticity even further with the introduction of Capacity Blocks and Capacity Dashboard — two new capabilities that bring additional clarity to GPU capacity management and give users full confidence in the compute they rely on.